Task 4.3: Modeling facility

Research partners: University of Lugano (USI), Swiss Federal Institute of Technology in Zurich (ETHZ), Swiss National Supercomputing Centre (CSCS)

Research objectives
Mathematical Modeling and numerical simulation allow for the virtual understanding, analysis, and optimization of almost all applications in geo-energy and hydropower. However, the extremely high complexity of the considered coupled physical systems (flow in fracture networks and porous media, hydro-thermal-mechanics, fluid structure interaction, wear, etc.)  requires simulation tools which are tailored specifically to geothermal and hydropower applications. It is the goal of the central modeling facility to integrate and provide know-how in state-of-the-art simulation and optimization techniques and to develop new scientific simulation software within the SCCER-SoE. This is done in  cooperation with the CSCS at ETHZ, so that the developed code can be optimized for recent and upcoming supercomputers at CSCS.

Current projects

This is a sub-project  of the joint project "Modelling permeability and stimulation for deep heat mining” within the framework of the NRP NRP 70 “Energy Turnaround” (see task 1.2). In the numerical simulation of stimulation processes for heat extraction, hydro-mechanical and thermal effects have to be considered. This requires not only specially adapted techniques for the discretization of the modeling partial differential equations, but also fast and stable solution methods for the arising non-linear systems. The resulting simulation tools has to be able to deal  with permeability changes in response to shearing on rough fracture surfaces in complex fracture networks. Due to the huge size of the modeled geometries (i.e. deep reservoirs), massively parallel methods (domain decomposition and multigrid) are needed  In this project we develop and implement of multigrid methods for frictional contact problems – including rough surfaces, whic eventually will allow for the fast simulation of deep heat mining on the large machines at CSCS.

For the numerical simulation of multi-physics problems in geo- and hydro science, simulation tools are required which allow for (hydro-thermal-mechanics, fluid-structure interaction, etc.) which allow for the coupling of different physical quantities (fluid velocity, temperature, displacements, stresses, etc.) along interfaces or within volumes. In this project, we develop transfer operators which are based on a variational approach, i.e. a diskrete L2-projection. In contrast to standard approaches, where interpolation is used, the variational transfer in general is more stable and therefore preferable. The goal of this sub-project is the development of a coupling library, which then can be used within the different software frameworks within the SCCER-SoE.

A central aspect in of numerical simulation in geo- and hydro science is the solution of large non-linear equation systems. Usually, they arise from the discretization of partial differential equations, modeling e.g. porous media, hydro-thermal mechanics, fluid structure interaction, flow in fracture networks, etc. Within this sub-project, we develop the new software library PASSO, which is designed as a stand alone solution for the massively parallel solution of non-linear problems. PASSO will in particular allow for the handling of non-smooth effects, such as friction or contact. The library will then be made available to the SCCER-SoE.

The numerical modeling of fluid structure interaction shows up in many applications - for example in hydro-sciences in the modeling of water turbines. In order to get a reasonable fast “time to solution” for numerical simulations, efficient and parallel solution methods are needed. The development of parallel methods for fluid-structure-interaction is far from trivial, as possibly different discretizations and different physical models (solid mechanics and fluid mechanics) have to be coupled. In this project, we have developed a new coupled discretization scheme based on Finite Volume Methods for the fluid and Finite Element Methods for the solid and have moreover developed and implemented a fast parallel solution method -based on overlapping domain decomposition approaches- which shows excellent scalability properties in our runs at CSCS.

This is a joint project with Siemens (Germany). The modeling of fatigue is an important aspect in the design of modern machine parts. In this project, we have developed a a new approach for quantifying the probability of crack initiation due to surface driven low-cycle fatigue (LCF).The devloped approach is based on failure- time processes and can be applied in a post-process to finite element simulations modeling the mechanical behavior of the machine part under consideration. Inhomogeneous stress fields and size effects are taken into account.